Effectiveness, Cost Savings of AI to Screen for Pediatric Diabetic Retinopathy

Despite recommendations to the contrary, screening rates for diabetic retinopathy among pediatric patients with type 1 and 2 diabetes remain low. Point-of-care screening is available for diabetic retinopathy screening, but its cost-effectiveness compared with standard screening by an eye care professional (ECP) is unclear. According to a recent study, the use of artificial intelligence (AI) in diabetic retinopathy screening among pediatric patients with type 1 and 2 diabetes was effective and resulted in cost savings.

Data collection spanning 1994 through 2019 included out-of-pocket costs for autonomous AI screening, ophthalmology visits, and treating diabetic retinopathy; probability of standard retinal examination receipt; relative screening odds; and the sensitivity, specificity, and diagnosability of the ECP versus autonomous AI screenings. The main outcomes were patient costs or savings, per mean patient payment for diabetic retinopathy screening examination, as well as cost-effectiveness, per costs or savings correlated with the number of true-positive results that diabetic retinopathy screening yielded.

In standard ophthalmologic screening performed by an ECP, the expected true-positive proportions for type 1 and 2 diabetes were 0.006 and 0.01, respectively; for autonomous AI, they were 0.03 and 0.04, respectively. With a base case scenario of 20% adherence, autonomous AI use resulted in a higher mean payment than conventional ECP screening for both patients with type 1 ($8.52 vs. $7.91) and type 2 ($10.85 vs. $8.20) diabetes. But with an adherence rate of at least 23%, autonomous AI was the preferred screening strategy.

“These results suggest that point-of-care diabetic retinopathy screening using autonomous AI systems is effective and cost saving for children with diabetes and their caregivers at recommended adherence rates,” the study authors summarized.